An arrhythmia is an abnormal heart beat pattern. One example of arrhythmia is bradycardia wherein the heart beats at an abnormally slow rate or wherein significant pauses occur between consecutive beats. Other examples of arrhythmia include tachyarrhythmias wherein the heart beats at an abnormally fast rate. With an atrial tachyarrhythmia, such as atrial tachycardia (AT), the atria of the heart beat abnormally fast. With a ventricular tachyarrhythmia, such as ventricular tachycardia (VT), the ventricles of the heart beat abnormally fast. Though often unpleasant for the patient, a tachycardia is typically not fatal. However, some tachycardias, particularly ventricular tachycardia, can trigger ventricular fibrillation (VF) wherein the heart beats chaotically such that there is little or no net flow of blood from the heart to the brain and other organs. VF, if not terminated, is fatal. Hence, it is highly desirable for implantable medical devices, such as pacemaker or ICDs (herein generally referred to as a pacer/ICD) to detect arrhythmias, particularly ventricular tachyarrhythmias, so that appropriate therapy can be automatically delivered by the device.
To detect arrhythmias, the pacer/ICD senses electrical cardiac signals within the heart of the patient using one or more implanted electrodes. The cardiac signals are sensed within the device by one or more sense amplifiers and then filtered by various filters configured so as to extract signals of interest, such as signals indicative of bradycardia or tachycardia or other arrhythmias. To this end, state-of-the-art pacer/ICD's are often provided with a wideband filter and two narrow bandwidth filters. The wideband filter eliminates low and high frequency noise but otherwise retains all features of the cardiac signals indicative of actual electrical events within the heart of the patient. That is, the wideband filter retains P-waves, R-waves and T-waves, whether occurring at normal heart rates, excessively low rates, or excessively high rates. The P-wave is the portion of an electrical cardiac signal that is representative of the electrical depolarization of the atria and is thus also representative of the physical contraction of the atria. The R-wave—which is a part of a QRS complex—is the portion of an electrical cardiac signal that is representative of the electrical depolarization of the ventricles and is thus also representative of the physical contraction of the ventricles. The T-wave is the portion of an electrical cardiac signal that is representative of the electrical repolarization of the ventricles. Note that the repolarization of the atria typically generates electrical signals that are too weak to be detected and hence atrial repolarization events are not typically detected. Hence, within the wideband cardiac signals, the P-wave is typically followed by the R-wave, which is then followed by the T-wave. Note, however, that the wideband filter also retains signals associated with any chaotic or random beating of the chambers of the heart, particularly signals associated with VF, which may not be easily categorized as having discrete P-waves, R-waves or T-waves. Also, note that, P-waves, R-waves and T-waves are also features of a surface electrocardiogram (EKG), though the corresponding features of the EKG often differ in shape and magnitude from those of the IEGM.
FIG. 1 provides a stylized illustration of a cardiac signal 2 corresponding to a single heartbeat, particularly illustrating the P-wave 4, R-wave 6, and the T-wave 8. In practice, the relative magnitudes of the various events can differ significantly. In some cases, the T-wave may be as large as or larger than the R-wave. Accordingly, it can be difficult to obtain an accurate measure of the ventricular rate from the wideband-filtered signals and so it can be difficult to reliably detect either bradycardia or tachycardia from the wideband-filtered signals. Hence, the specialized narrowband filters have been developed. Initially, a narrowband “bradycardia filter” was provided within pacemakers that passed (or “retained”) only R-waves for the purposes of detecting bradycardia. If the rate at which R-waves appear in the filtered signal is below a lower rate threshold, or if no R-waves are present at all in the filtered signal, then the patient is likely suffering an episode of bradycardia, and appropriate therapy is delivered, such as demand-based pacing. Although effective for detecting bradycardia, the filter also eliminates R-waves associated with VF, i.e. the bradycardia filter also filtered out V-fib waves. ICDs need to reliably detect VF for the purposes of delivering defibrillation shocks. Hence, a narrowband “tachycardia” filter was also developed that had a wider passband for the purposes of also detecting R-waves or V-fib waves associated with VF/VT. If the rate at which R-waves appear in the tachycardia-filtered signal is above a VT threshold, then the patient is likely suffering an episode of VT, and appropriate therapy can be delivered, such as antitachycardia pacing (ATP). If the rate exceeds a higher VF threshold, or if V-fib waves are detected, then the patient is likely suffering an episode of VF, and defibrillation shocks are delivered.
Accordingly, many state-of-the art pacer/ICDs now include both a bradycardia filter and a tachycardia filter. Advantageously, because T-waves are filtered out by the bradycardia filter, the sensitivity of the bradycardia filter can be set quite high so as to permit detection of even very low amplitude R-waves. The high sensitivity of the bradycardia filter thus substantially eliminates the risk of any possible undersensing of the R-waves (or at least any significant undersensing of relatively low rate R-waves.) Herein, “undersensing” refers to the failure to detect events of interest that are actually present within the raw cardiac signals. Meanwhile, the elimination of T-waves means that there is substantially no risk of “oversensing” when using the bradycardia filter. Herein, “oversensing” refers to the erroneous detection of an event not actually present in the raw cardiac signal, such as the detection of R-waves that are not in fact present. Oversensing typically arises when one event is misidentified as another, as may occur, e.g., if a T-wave is improperly identified as an R-wave. As can be appreciated, T-wave oversensing is a significant concern since misidentification of T-waves as R-waves can result in significant miscalculation of the true heart rate within the patient, causing therapy to be delivered when not warranted or potentially causing therapy to be withheld even when needed. Insofar as bradycardia is concerned, T-wave oversensing might result in a failure to detect bradycardia since misidentification of T-waves as R-waves would result in a significantly higher heart rate being detected than actually occurring within the patient. As noted, the bradycardia filter is configured to substantially eliminate all T-waves so that T-wave oversensing is not a concern on the bradycardia channel. Hence, the state-of-the art pacer/ICD can reliably use the bradycardia filter to detect bradycardia.
FIG. 2 illustrates the operation of the bradycardia filter during normal sinus rhythm. A first graph 10 of the figure illustrates the output of the wideband filter, particularly highlighting that portion of the filtered signal corresponding to the location of T-waves 12. Note that, in the figure, T-waves corresponding to numerous heartbeats are shown superimposed over one another. The vertical axis of the graph illustrates the magnitude, in arbitrary units, relative to an iso-electric baseline. The horizontal axis illustrates the time delay in milliseconds (ms) from the preceding R-wave. A portion of each preceding R-wave appears within the graph around time: 0 ms. A portion of each subsequent R-wave appears within the graph as well, beginning at about 500 ms. Between them, T-waves are clearly seen.
Meanwhile, a second graph 14 of FIG. 2 illustrates the output of the bradycardia filter. Again, cardiac signals corresponding to numerous heartbeats are shown superimposed over one another. As can be seen, T-waves are completely eliminated within the bradycardia signals, leaving only those portions corresponding to R-waves (again seen at the beginning and at the end of the highlighted section of the cardiac signal.) By completely eliminating the T-wave, the ventricular rate can be easily and accurately measured based solely on the R-waves (at least at relatively low heart rates), and hence bradycardia can be reliably detected from the filtered signals.
However, unlike the bradycardia filter, which fully eliminates T-waves, the tachycardia filter retains T-waves. This is due to the fact that the frequencies associated with the V-fib waves of interest are also associated with T-waves, and hence the filter cannot eliminate all T-waves while still retaining the V-fib waves. As such, the sensitivity of the tachycardia filter must be set so as to detect high rate R-waves and V-fib waves while eliminating T-waves. This is difficult, at best, since the relative magnitudes of the R-waves, V-fib waves and T-waves may change significantly over time within the patient, perhaps due to the use of medications or due to physiological or anatomical changes in the heart brought on by medical conditions, such as cardiac ischemia, myocardial infarctions, congestive heart failure, etc. Moreover, as already noted, T-waves can sometimes have a magnitude that equals or exceeds that of the R-wave. Hence, T-wave oversensing is a significant problem within the tachycardia-filtered signals.
FIG. 3 illustrates the operation of a tachycardia filter during VT/VF. A first graph 16 illustrates the output of the wideband filter, again particularly highlighting that portion of the filtered signal corresponding to the location of T-waves 18. Note that, as in the previous figure, T-waves corresponding to numerous heartbeats are shown superimposed over one another. The vertical axis of the graph again illustrates magnitude relative to an iso-electric baseline. The horizontal axis again illustrates the time delay from the preceding R-wave. A portion of each preceding R-wave appears within the graph around time: 0 ms-10 ms. A portion of each subsequent R-wave appears within the graph as well, beginning at about 60 ms. Between them, T-waves 18 are clearly seen. (As a result of variations in R-R intervals occurring during VT, the R-waves and T-waves from different heartbeats are not aligned with one another within the graph, as was the case in FIG. 1.)
Meanwhile, a second graph 20 of FIG.2 illustrates the output of the tachycardia filter. Again, cardiac signals corresponding to numerous heartbeats are shown superimposed over one another. As can be seen, T-waves 22 are not completely eliminated within the tachycardia-filtered signals, leaving signals that might be misidentified as R-waves, particularly if the sensitivity of the tachycardia filter is set too high. That is, T-wave oversensing might occur. Without complete elimination of the T-wave, the ventricular rate cannot be accurately and reliably measured based solely on the output of the tachycardia filter (at least at the rates associated with VT/VF), and hence problems arise in the detection of VT/VF or other forms of ventricular tachyarrhythmia. Failure to properly detect VT/VF when it is present can result in a failure to deliver appropriate therapy. False detection of VT/VF when it is not present can result in delivery of inappropriate therapy. As can be appreciated, both situations are of significant concern.
In view of the problems arising when using a narrowband tachycardia filter, it is highly desirable to provide improved techniques for reliably detecting VT/VF that may be performed by a pacer/ICD. It is to this end that various aspects of the invention are generally directed. It is particularly desirable to provide improved techniques that do not require replacement or elimination of existing tachycardia filters, but that instead achieve improved VT/VF detection when using otherwise conventional tachycardia filters. It is to this end that particular aspects of the invention are directed.
Still further aspects of the invention are directed to providing improved techniques for detecting and eliminating T-wave oversensing, even in the absence of any arrhythmia. Heretofore, at least some techniques for addressing T-wave oversensing have been directed to providing blanking intervals synchronized with the expected location of the T-wave. See, for example, U.S. Pat. No. 6,862,471 to McClure, et al., entitled “Method and Apparatus for Blanking T-Waves from Combipolar Atrial Cardiac Signals based on Expected T-Wave Locations.” It would be desirable to provide techniques for detecting and eliminating T-wave oversensing that do not necessarily require the use of blanking intervals, and various aspects of the invention are directed to that end as well.